ľ¹Ï¸£ÀûÓ°ÊÓ

Accelerating gravitational-wave inference with machine learning

Bayesian inference underpins modern astronomical discovery, yet it is often computationally intensive. This challenge is especially pronounced in gravitational-wave astrophysics, where the growing number of observed compact binary mergers demands faster and more scalable analysis methods.

In this talk, I will show how machine learning can accelerate gravitational-wave inference. I will focus on two general-purpose approaches that use normalizing flows—a class of flexible models for learning probability distributions—as drop-in replacements for existing inference algorithms. These methods improve efficiency while remaining compatible with established analysis frameworks. I will present results demonstrating significant improvements over current techniques and highlight new types of analyses made possible by these methods. I will conclude with a discussion of their advantages, limitations, and considerations for broader adoption.